I am Sarah Cliffe, the executive editor of the “Harvard Business Review.” Before executives make high-risk decisions, they need to think first about how they're going to decide. A lot of process questions do come up, and one of the important ones is what decision support tool to use. There are a lot of good decision support tools, but many of the more traditional ones are not helpful at all when a business faces high levels of uncertainty, which -- let's face it -- many of us do today.
This decision tree, which features hypothetical decisions that McDonald's executives might need to make, was developed by Hugh Courtney, Dan Lovallo, and Carmina Clarke, and is described in the November 2013 issue of “HBR.” Their methodology breaks down the questions you need to ask yourself when faced with a major decision. It should lead you to the right decision support tool or set of tools.
The tree relies on three key questions, two of which we'll outline here. First, do I know what it will take to succeed? In other words, do I understand my full causal model for success? And second, can I predict the range and likelihood of different possible outcomes of the decision? We're going to work our way through those questions, moving from a very straightforward context to a very uncertain one. We'll start with a simple decision. McDonald's needs to decide where to locate new US restaurants.
In this case, is the full causal model known? It absolutely is. The company knows the variables that matter for success, things like local demographics, foot traffic patterns, real estate prices, and so on. In addition, it has terrific data sources on all of those variables, and it has a well-calibrated, well-tested revenue and cost model. In addition, the decision outcomes are known, partly because they have such good information. McDonald's can predict with some precision how well a location will perform because it's opened so many US stores in the past.
In this really familiar context, executives can use standard, conventional capital budgeting tools, things like standard discounted cash flow, to make a clear go/no go decision. Now it's say McDonald's is deciding whether to introduce a new sandwich into the US market. They still have relevant data about demographics, foot traffic, and so forth. So they do know what it will take to succeed.
But there's some uncertainty about the outcome of introducing a sandwich. Executives don't know for sure what the demand will be, nor do they know what impact it will have on the sale of complementary products. However, some market research in different parts of the country can solve that problem, will give them a range of possible outcomes, and some probabilities. They can then use a tool like Monte Carlo simulation or a real options analysis. These are quantitative multiple scenario tools to make a decision about whether to introduce the sandwich and where.
Now let's ratchet the uncertainty up a notch. What if McDonald's is entering an emerging market for the first time? Executive still know what it takes to succeed. The variables, like demographics, and the basic cost and revenue models, are more or less the same. However, the decision outcomes aren't known, since this is the company's first entrance in this locale. So predicting outcomes using market research and statistical analysis, as we did with the sandwich introduction, would be difficult.
In this situation, McDonald's can use qualitative scenario analysis to get a better sense of possible outcomes. It can build scenarios covering a range of customer acceptance rates and a range of supplier costs and reliability levels. Executives should then supplement those scenarios with case-based decision analysis of similar business situations, perhaps looking at outcomes from their own entries into other developing markets.
Now let's say McDonald's wants to enter a new line of business with a completely new business model, like consulting on food service process improvements. Since this is a brand new situation, executives probably can't define their full causal model. However, they probably can to define the range of possible outcomes if they tap into the right information sources, for example, by studying companies who have more experience with this business model.
In this case, McDonald's best bet is to go straight to case-based decision making, to find multiple analogies, then develop a rigorous comparison between itself and those reference cases. Now for the doozy, where the answer to both questions is no. What if McDonald's needs to respond to concern about obesity in the US and a backlash over the fast food industry's role in that epidemic?
The backlash has the potential to fundamentally rewrite the rules for the fast food industry and to make their historical data obsolete. So McDonald's really doesn't know what success will look like, nor can the company foresee a range of possible outcomes to moves they might make, since they can't accurately forecast future lawsuits, medical research, competitor moves, or consumer reactions.
When faced with this very high level of uncertainty, McDonald should once again rely on case-based decision analysis. Relevant comparisons might include beverage companies' attempts to reposition themselves as healthy or safe, or the firearms industries efforts to influence regulation and legislation. Every high stakes decision has its own challenges, of course. But this methodology can help you make smarter decisions about what tools to use when making them.
You can learn more about how to use decision support tools in the article “Deciding How to Decide” in the November 2013 issue of “ HBR.” We hope you'll enjoy it.